Voice biometric passwords are used as a form of speaker verification and are available on some systems and may grow in ubiquity in the near term. Such passwords are based upon identifying features in speech, which when taken in combination uniquely identify an individual with a certain probability.
Noise reduction algorithms may be used to reduce the effects of background noise on both collection of samples to form models representing features in speech and in use of the models to identify speakers. Such noise reduction algorithms may also affect system performance.